The answer is… its complicated, so you approximate. A good way of approximating is getting a dataset together and putting together a good model that helps explain that dataset. Doing the perfect Bayesian update in the real world is usually worse than nontrivial—its basically impossible.
The answer is… its complicated, so you approximate. A good way of approximating is getting a dataset together and putting together a good model that helps explain that dataset. Doing the perfect Bayesian update in the real world is usually worse than nontrivial—its basically impossible.